emilstenstrom / syntaxnet-api Goto Github PK

A Dockerfile that helps you get Syntaxnet DRAGNN working, with a json api powered by falcon.

Here's how you get everything running:

1. Download the pretrained model files

   Download this file from Google Drive: https://drive.google.com/uc?id=0BxpbZGYVZsEeSFdrUnBNMUp1YzQ&export=download

   Unzip it into a directory where you want your model stored.

2. Start the server

   Save this to a file called rebuild-and-run.sh, changing things to fit your environment (see below):

   #!/bin/bash
   docker build --rm -t syntaxnet-api -f Dockerfile .
   
   # :ro at the end of -v makes volume read-only
   docker run --rm -p 8000:80 -v ~/Projects/conll17-dragnn/:/usr/local/tfmodels:ro -e LANGUAGE="Swedish" syntaxnet-api

   • syntaxnet-api is the name of the container
   • 8000 is the port where the API will be exposed
   • ~/Projects/conll17-dragnn/ is the directory where you unziped the model files
   • /usr/local/tfmodels is the directory inside the container (don't change this)
   • -e LANGUAGE="Swedish" is the name of the directory where the files for your language resides

   Make it runnable: chmod +x rebuild-and-run.sh

   Now run it: ./rebuild-and-run.sh

   This will start a falcon server that runs on port 8000.

3. Use the API

   The API can be used from any program that can send HTTP requests and parse JSON. Here's how you would send data to it using plain curl from the command line:

   curl -i -XPOST --data-urlencode "data=Detta är en mening." http://localhost:8000 && echo ""